Corrections

MEDICAL SCIENCES Correction for “Regulation of bone remodeling by vasopressin New, Alberta Zallone, and Mone Zaidi, which appeared in issue 46, explains the bone loss in hyponatremia,” by Roberto Tamma, November 12, 2013, of Proc Natl Acad Sci USA (110:18644–18649; Li Sun, Concetta Cuscito, Ping Lu, Michelangelo Corcelli, first published October 28, 2013; 10.1073/pnas.1318257110). Jianhua Li, Graziana Colaianni, Surinder S. Moonga, Adriana The authors note that Fig. 1 appeared incorrectly. The cor- Di Benedetto, Maria Grano, Silvia Colucci, Tony Yuen, Maria I. rected figure and its legend appear below.

Fig. 1. Bone cells express Avprs. Immunofluorescence micrographs (A) and Western immunoblotting (B) show the expression of Avpr1α in osteoblasts and osteoclasts, and as a function of osteoblast (mineralization) and osteoclast (with Rankl) differentiation. The expression of Avp (ligand) and Avpr1α (receptor) in osteoblasts is regulated by 17β-estradiol, as determined by quantitative PCR (C) and Western immunoblotting (D). (Magnification: A,63×.) Because Avp is a small peptide, its precursor neurophysin II is measured. Statistics: Student t test, P values shown compared with 0 h. Stimulation of Erk phosphorylation − (p-Erk) as a function of total Erk (t-Erk) by Avp (10 8 M) in osteoclast precursors (preosteoclasts), osteoclasts (OC), and osteoblasts establishes functionality of − the Avpr1α in the presence or absence of the receptor inhibitor SR49059 (10 8 M) (E). Western immunoblotting showing the expression of Avpr2 in pre- −/− osteoclasts, OCs (F), and osteoblasts (G) isolated from Avpr1α mice, as well as in MC3T3.E1 osteoblast precursors (G). Functionality of Avpr2 was confirmed −/− by the demonstration that cells from Avpr1α mice remained responsive to AVP in reducing the expression of osteoblast differentiation , namely Runx2, Osx, Bsp, Atf4, Opn, and (quantitative PCR, P values shown) (H). Only relevant bands from Western blots are shown, with gaps introduced where empty lanes are excised to conserve space.

www.pnas.org/cgi/doi/10.1073/pnas.1415306111

14002–14003 | PNAS | September 23, 2014 | vol. 111 | no. 38 www.pnas.org Downloaded by guest on September 27, 2021 PHARMACOLOGY PHYSIOLOGY Correction for “Arrhythmias, elicited by catecholamines and Correction for “Hypothalamic prolyl endopeptidase (PREP) serotonin, vanish in human chronic atrial fibrillation,” by Torsten regulates pancreatic insulin and glucagon secretion in mice,” by Christ, Nadiia Rozmaritsa, Andreas Engel, Emanuel Berk, Michael Jung Dae Kim, Chitoku Toda, Giuseppe D’Agostino, Caroline J. Knaut, Katharina Metzner, Manuel Canteras, Ursula Ravens, Zeiss, Ralph J. DiLeone, John D. Elsworth, Richard G. Kibbey, and Alberto Kaumann, which appeared in issue 30, July 29, 2014, Owen Chan, Brandon K. Harvey, Christopher T. Richie, Mari of Proc Natl Acad Sci USA (111:11193–11198; first published Savolainen, Timo Myöhännen, Jin Kwon Jeong, and Sabrina Diano, July 14, 2014; 10.1073/pnas.1324132111). which appeared in issue 32, August 12, 2014, of Proc Natl Acad The authors note that an additional affiliation should be listed Sci USA (111:11876–11881; first published July 28, 2014; 10.1073/ for Torsten Christ and Alberto Kaumann. The new affiliation pnas.1406000111). should appear as Department of Experimental Pharmacology and The authors note that the author name Timo Myöhännen Toxicology, University Medical Center Hamburg-Eppendorf. The should instead appear as Timo Myöhänen. The corrected author corrected author and affiliation lines appear below. The online line appears below. The online version has been corrected. version has been corrected. Jung Dae Kim, Chitoku Toda, Giuseppe D’Agostino, Torsten Christa,b, Nadiia Rozmaritsaa, Andreas Engela, Caroline J. Zeiss, Ralph J. DiLeone, John D. Elsworth, Emanuel Berka, Michael Knautc, Katharina Metznera, Richard G. Kibbey, Owen Chan, Brandon K. Harvey, Manuel Canterasd, Ursula Ravensa, and Christopher T. Richie, Mari Savolainen, Timo Myöhänen, Alberto Kaumannb,e Jin Kwon Jeong, and Sabrina Diano

Departments of aPharmacology and Toxicology and cHeart Surgery, www.pnas.org/cgi/doi/10.1073/pnas.1416210111 Dresden University of Technology, 01307 Dresden, Germany; bDepartment of Experimental Pharmacology and Toxicology, University Medical Center Hamburg-Eppendorf, D-20246 Hamburg, Germany; and Departments of dBiostatistics and ePharmacology, University of Murcia, 30100 Murcia, Spain

www.pnas.org/cgi/doi/10.1073/pnas.1416216111 CORRECTIONS

PNAS | September 23, 2014 | vol. 111 | no. 38 | 14003 Downloaded by guest on September 27, 2021 Arrhythmias, elicited by catecholamines and serotonin, vanish in human chronic atrial fibrillation

Torsten Christa,b,1,2,3, Nadiia Rozmaritsaa,1,4, Andreas Engela,5, Emanuel Berka,6, Michael Knautc, Katharina Metznera,7, Manuel Canterasd, Ursula Ravensa, and Alberto Kaumannb,e,2

Departments of aPharmacology and Toxicology and cHeart Surgery, Dresden University of Technology, 01307 Dresden, Germany; bDepartment of Experimental Pharmacology and Toxicology, University Medical Center Hamburg-Eppendorf, D-20246 Hamburg, Germany; and Departments of dBiostatistics and ePharmacology, University of Murcia, 30100 Murcia, Spain

Edited* by Lutz Birnbaumer, National Institute of Environmental Health Sciences, Research Triangle Park, NC, and approved June 23, 2014 (received for review December 30, 2013) Atrial fibrillation (AF) is the most common heart rhythm disorder. Increased propensity to generate spontaneous impulses is as- Transient postoperative AF can be elicited by high sympathetic sumed to initiate and/or maintainAFinhumans.Arrhythmiasmay nervous system activity. Catecholamines and serotonin cause ar- develop through spontaneous impulse generation within individual rhythmias in atrial trabeculae from patients with sinus rhythm (SR), myocytes and/or reentry around nonexcitable tissue. The traveling but whether these arrhythmias occur in patients with chronic AF is electrical impulse of a premature atrial beat can encounter areas of unknown. We compared the incidence of arrhythmic contractions refractoriness, return to its origin in a retrograde way, and through caused by norepinephrine, epinephrine, serotonin, and forskolin in reentry initiate and maintain AF. Spontaneous impulse generation atrial trabeculae from patients with SR and patients with AF. In the could be related to increased activity of PKA and/or CaMKII, with + patients with AF, arrhythmias were markedly reduced for the subsequent uncoordinated release of Ca2 from the sarcoplasmic + agonists and abolished for forskolin, whereas maximum inotropic reticulum. Such a concept is attractive, because Ca2 released from + + responses were markedly blunted only for serotonin. Serotonin and the “leaky” would activate the Na -Ca2 2+ + forskolin produced spontaneous diastolic Ca releases in atrial myo- exchanger to extrude Ca2 and to produce a arrhythmogenic cytes from the patients with SR that were abolished or reduced 2+ depolarizing current, thereby explaining both the contractile in myocytes from the patients with AF. For matching L-type Ca - dysfunction and the high recurrence rate (11–13). current (ICa,L) responses, serotonin required and produced ∼100-fold 2+ If arrhythmias were initiated and maintained by increased less cAMP/PKA at the Ca channel domain compared with the cat- activity of PKA and/or CaMKII, then interventions known to echolamines and forskolin. Norepinephrine-evoked ICa,L responses + were decreased by inhibition of Ca2 /-dependent II (CaMKII) in myocytes from patients with SR, but not in those from Significance patients with AF. Agonist-evoked phosphorylation by CaMKII at phospholamban (Thr-17), but not of ryanodine2 (Ser-2814), was re- Catecholamines and serotonin elicit arrhythmias in atrial tra- 2+ duced in trabeculae from patients with AF. The decreased CaMKII beculae and arrhythmogenic diastolic Ca releases in myocytes activity may contribute to the blunting of agonist-evoked arrhyth- from patients with normal sinoatrial rhythm (SR). Arrhythmic mias in the atrial myocardium of patients with AF. events are greatly blunted in the myocardium of patients with chronic atrial fibrillation (AF). The mediation by cyclic AMP- + dependent kinase of L-type Ca2 -current (I ) responses trial fibrillation (AF) is the most common cardiac arrhythmia, Ca,L to agonists, , and protein phosphorylation are preserved associated with increased risk of death, congestive heart fail- A in AF. The vanishing of agonist-evoked arrhythmias in AF is as- ure, and stroke (1). AF is characterized by an irregular, often rapid sociated with the disappearance of an ICa,L response, facilitated by . Atria contract with reduced force, thereby favoring 2+

Ca /calmodulin-dependent kinase II (CaMKII), in myocytes from PHARMACOLOGY thrombus formation (2). AF occurs in several cardiac diseases, and patients with SR, and decreased phosphorylation of phospho- its incidence is higher in woman than in men, particularly in those SI Appendix lamban, but not ryanodine, by this enzyme. The disappearance of with valvular heart disease ( , Table S1). Chronic AF arrhythmias and associated reduction of CaMKII functions causes structural and electrical remodeling, as well as enlarged atria questions the use of CaMKII inhibitors in AF, as has been pro- SI Appendix ( ,TableS1),whichinturncontributestomaintainAF. posed by other groups. Atrial contractility is reduced in isolated atrial tissues (3) obtained from patients with chronic AF, attributed to a marked Author contributions: T.C., N.R., and A.K. designed research; T.C., N.R., A.E., E.B., M.K., 2+ decrease in L-type Ca current (ICa,L) (3, 4). The inotropic K.M., and A.K. performed research; T.C., N.R., A.E., E.B., K.M., M.C., U.R., and A.K. ana- lyzed data; and A.K. wrote the paper. responses to the agonist for β1-adrenergic receptors (β1ARs) and The authors declare no conflict of interest. β2-adrenergic receptors (β2ARs), isoproterenol (ISO), but not the density of βARs and G , are decreased in AF (3); *This Direct Submission article had a prearranged editor. 1 however, whether the function of coexisting β1ARs and/or β2ARs T.C. and N.R. contributed equally to this work. 2 is perturbed is unknown. Activation of human atrial β1ARs, To whom correspondence may be addressed. Email: [email protected] or [email protected]. 3 β2ARs, and serotonin [5-hydroxytryptamine (5-HT)] 5-HT4 Present address: Department of Experimental Pharmacology and Toxicology, University receptors hastens relaxation through phosphorylation of phos- of Hamburg, 20246 Hamburg, Germany. pholamban (PLB) (5–7) by cAMP-dependent protein kinase 4Present address: Division of Cardiovascular Medicine, University of Oxford, Oxford OX3 + (PKA) and Ca2 /calmodulin-dependent kinase II (CaMKII) and 9DU, United Kingdom. produces arrhythmias (8, 9) in atrial trabeculae in patients with 5Present address: Institute for Legal Medicine, Medical Faculty, Technical University Dresden, 01307 Dresden, Germany. sinus rhythm (SR). Catecholamines and 5-HT have been pro- 6Present address: Department of Internal Medicine II, Municipal Hospital Dresden- posed to initiate AF (8, 9). The relevance of these in vitro Friedrichstadt, 01067 Dresden, Germany. arrhythmias is corroborated by the clinical finding that high 7Present address: Rudolf Boehm Institute of Pharmacology and Toxicology, University of sympathetic nervous system activity during and after cardiac Leipzig, 04107 Leipzig, Germany. surgery causes premature beats and transient postoperative AF This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. in approximately one-third of patients (10). 1073/pnas.1324132111/-/DCSupplemental. www.pnas.org/cgi/doi/10.1073/pnas.1324132111 PNAS | July 29, 2014 | vol. 111 | no. 30 | 11193–11198 stimulate both would be expected to evoke more arrhyth- inotropic potency to FSK was reduced by sevenfold. FSK failed mias in patients with AF. However, in vitro induction of arrhythmias to elicit arrhythmias in AF (Fig. 1). Because ICa,L responses to and activation of PKA and CaMKII by catecholamines and 5-HT 5-HT were similar to the responses of catecholamines and FSK were not assessed in tissues from patients with AF. Thus, we com- in myocytes from patients with SR and patients with AF (see pared the effects of endogenous agonists on force and arrhythmias below), the blunted arrhythmias in AF suggest a decrease in the 2+ 2+ in intact trabeculae, as well as the ICa,L,Ca transients (CaTs), and cytoplasmic arrhythmogenic Ca surges in AF produced by + diastolic Ca2 release, in atrial myocytes obtained from patients with 5-HT and FSK. CaTs in the absence of 5-HT and FSK were SR and patients with AF. Functional measurements, including re- monophasic. In myocytes from patients with SR, both 5-HT and laxation, were supplemented by Western blot analysis of relevant FSK increased CaTs, which became biphasic and often asso- + targets of PKA and CaMKII. Chronic AF caused a marked de- ciated with spontaneous diastolic Ca2 releases (SDCRs) (Fig. crease in agonist-evoked arrhythmias, associated with a decrease in 2, and SI Appendix,Figs.S1–S3). Interestingly, ISO (1 μM) some CaMKII-catalyzed functions but unchanged PKA functions. caused similar results with biphasic CaTs and SDCRs (SI Appendix,Fig.S4). In AF, the CaTs generated by 5-HT and Results FSK were reduced and lost their biphasic shape (SI Appendix, In AF Arrhythmias Elicited by Catecholamines, 5-HT, and Forskolin Are Figs. S1–S3).SDCRswereabsent(5-HT)orreduced(FSK)in Markedly Reduced, and Inotropic Responses to 5-HT Are Selectively AF (Fig. 2), in line with the failure of 5-HT and FSK to Depressed. The maximum inotropic response to 5-HT, but not to elicit arrhythmias. norepinephrine (NE), epinephrine (EPI), or forskolin (FSK), was markedly reduced in trabeculae of patients with AF. The inotropic Maximum ICa,L Responses to NE, EPI, 5-HT, and FSK Are Similarly Reduced in AF, but Control by PKA Is Preserved. potencies (−log EC50 values) of the four drugs were decreased (Fig. ICa,L is the initial 1andSI Appendix,TableS2).NE,EPI,5-HT,andFSKcaused step in electromechanical coupling, believed to contribute to concentration-dependent arrhythmic contractions with maximum increased contractile force and generation of arrhythmias (16). incidences of 17%, 14%, 4%, and 14%, respectively, in trabeculae of The maximum ICa,L responses to NE, EPI, 5-HT, and FSK were patients with SR. In trabeculae of patients with AF, the incidence of similar in patients with SR and similarly reduced by approxi- arrhythmic contractions (Fig. 1) was markedly reduced (≤2% for the mately one-third in patients with AF, but the potencies appeared agonists, abolished for FSK). Our results with 5-HT are in line with unchanged in AF (Fig. 3 and SI Appendix, Table S3), consistent a decrease in the arrhythmogenic electrophysiological effects of 5- with preserved function of the receptors and adenylyl cyclase HT reported in atrial myocytes from patients with AF compared (AC). The reduced ICa,L responses in myocytes from patients with AF could be related to reduced channel density or reduced with patients with SR (15). + PKA-catalyzed Ca2 channel phosphorylation and/or reduced Biphasic CaTs and Spontaneous Diastolic Ca2+ Releases Induced by access of cAMP to the channel domain. 5-HT and FSK in Myocytes from Patients with SR Are Blunted in To quantitatively assess the cAMP/PKA-dependent regulation of Patients with AF. In the patients with AF, the inotropic and ar- ICa,L by the agonists and FSK, we dialyzed the cardiomyocytes with rhythmic responses to 5-HT were markedly decreased, and the the nonhydrolyzable analog Rp-8-Br-cAMPS, which competes

Fig. 1. Arrhythmias, elicited by agonists and FSK, are markedly reduced and inotropic responses to 5-HT selectively depressed in AF. (A) Representative trac- ings of cumulative force-response curves to NE, EPI,

5-HT, and FSK, mediated through β1ARs, β2ARs, 5-HT4Rs, andAC.TheACactivatorFSKwasusedtoproducere- ceptor-independent accumulation of cAMP (14). A tra- beculum of a patient with SR (Upper) and a trabeculum of a patient with AF (Lower) were set up as pairs into thesameorganbath.(B) Representative arrhythmias elicited in trabeculae from patients with SR by 300 nM of each NE, EPI, 5-HT, and 10 μM FSK. (C) Inotropic data. Each data point represents mean values ± SEM. As reported previously (3), basal contractile force of atrial trabeculae was fourfold to fivefold smaller in patients with AF compared with those with SR. (D)Incidenceof arrhythmias. Number of experiments is given in trabec- ulae/patient. P values compare arrhythmias between patients with SR and patients with AF.

11194 | www.pnas.org/cgi/doi/10.1073/pnas.1324132111 Christ et al. + Fig. 2. Effects of 5-HT and FSK on spontaneous Ca2 release in cardiomyocytes from patients with SR and AF. (A) Examples of original CaTs traces before and after exposure to 5-HT (100 μM) in a cardiomyocyte, from a patient with SR (Upper) and a patient with AF (Lower). (Insets) Enlarged part of the traces. Arrows indicate spontaneous Ca2+ release. (B) Calculated fre- quency of spontaneous CaTs before and after 1 and 2 min of 5-HT application in cardiomyocytes from patients with SR and patients with AF, respectively. (C and D) Examples of original CaT traces and calcu- lated frequency of spontaneous CaTs before and after 1and2minofFSK(10μM) application in cardiomyocytes from patients with SR and patients with AF. *P < 0.05; **P < 0.01, 1 min vs. mean basal values; #P < 0.05, 2 min vs. basal values; all paired t tests. Numbers in the col- umns indicate myocytes/patients. with cAMP for binding to the regulatory unit of PKA (17). The Lusitropy and PKA-Catalyzed Phosphorylation of Proteins by the more cAMP produced, the greater the amount of Rp-8-Br- Agonists and FSK Are Not Different Between Patients with SR and Patients with AF. cAMPS needed to block PKA activation. ICa,L was not mod- The expression of proteins involved in lusitropy, ified by Rp-8-Br-cAMPS in patients with SR or those with AF PLB, troponin-I (Tn-I), and myosin binding protein C (Prot-C), (Fig. 4A) in the absence of agonists, inconsistent with regu- as well as SERCA2, did not differ between patients with SR and patients with AF (SI Appendix, Fig. S5). Basal phosphorylation lation of basal ICa,L by cAMP-activated PKA. Rp-8-Br-cAMPS caused concentration-dependent reductions in the I of PLB-Ser-16 was hardly detectable in trabeculae contracting Ca,L for 1 h in the organ bath used for the relaxation measurements responses to the agonists. The corresponding −log IC values 50 (SI Appendix, Figs. S6 and S7), suggesting that in the absence of of Rp-8-Br-cAMPS were not different for the agonists and agonists, this protein contributes little to relaxation. In contrast, FSK in myocytes from patients with SR and from patients with A basal phosphorylation of Tn-I-Ser23/24 and Prot-C-Ser-282 was AF (Fig. 4 ), demonstrating that the uniformly reduced re- detectable under these conditions, but did not differ between sponse of ICa,L to the agonists and FSK in AF is not related to patients with SR and patients with AF (SI Appendix,Figs.S6 reduced cAMP/PKA availability. Remarkably, however, the and S7). Our data are at variance with a report of increased − – log IC50 value of Rp-8-Br-cAMPS to reduce the 5-HT phosphorylation of PLB-Ser-16 and decreased phosphoryla- evoked ICa,L response was 2 log units lower than the corre- tion of Prot-C-Ser-282 in atria from patients with AF, frozen sponding –log IC values for NE, EPI, and FSK (Fig. 4A), 50 in the operating theater (11). Thus, we compared the phos- PHARMACOLOGY suggesting a 100-fold lower 5-HT–evoked cAMP production phorylation of PLB at Ser-16 in trabeculae frozen in the op- relative to ICa,L increases. erating theater with trabeculae from the same patient that was

Fig. 3. Similar reductions in the ICa,L response to NE, EPI, 5-HT, and FSK in AF. (A) Representative plots for ICa,L in response to the agonists (100 μM) and FSK (10 μM) in myocytes from patients with SR or AF. Dotted line indicates 0 current level. (B) Concentration-effect curves for the agonists and FSK. Red horizontal lines through the mid-

point of the curves are SEM of –logEC50s. Each data point represents mean values ± SEM in cells/patients. As reported

previously (4), basal ICa,L was decreased in myocytes from patients with AF compared with patients with SR.

Christ et al. PNAS | July 29, 2014 | vol. 111 | no. 30 | 11195 Fig. 4. Evidence for unchanged PKA function in AF. (A) Concentration-effect curves of Rp-8-Br-cAMPS for the ti- 2+ tration of PKA activity of ICa,L at the sarcolemmal Ca channel domain under the influence of 100 μM NE, 100 μM EPI, 100 μM 5-HT, and 10 μM FSK. The midpoint and red horizontal bars in the inhibitory curves for Rp-8-Br-cAMPS represent the concentration for half-maximal inhibition

(–logIC50 ± SEM) of the ICa,L responses. Open and closed symbols represent data from patients with SR and patients

with AF, respectively. Gray symbols represents basal ICa,L data in the absence of agonists and FSK. Each data point represents mean ± SEM. Total number of cells/patients in- vestigated from patients with SR vs. patients with AF were as follows: NE: 86/18 vs. 57/13; Epi: 103/24 vs. 53/12; 5-HT: 53/15 vs. 71/21; FSK: 39/8 vs. 35/8. C indicates values obtained from control experiments without Rp-8-Br-cAMPS. Note that Rp-8-Br-cAMPS was 2 log units (∼100-fold) more potent in

inhibiting the ICa,L response to 5-HT compared with the responses to the catecholamines and FSK. (B) Western blots of PLB-Ser-16, RyR2-Ser-2808, Tn-I-Ser-23/24, and Prot-C-Ser- 282 phosphorylation caused by agonists and FSK. Data in columns are from atria with four trabeculae of the same patient with SR (open columns) or patient with AF (black columns). *P < 0.05, **P < 0.01; ***P < 0.001 for the re- sponses to 5-HT compared with the responses to FSK. Phos- phorylation responses are shown as mean values ± SEM in arbitrary units (a.u.). CSQ, . NE, EPI, 5-HT, and FSK caused similar PKA-catalyzed phosphorylations in patients with SR or AF. Responses to 5-HT tended to be smaller than responses to NE, EPI, and FSK.

contracted for 1 h in an organ bath. We also measured the KN-92, reduced basal ICa,L in myocytes from patients with SR CaMKII-catalyzed phosphorylation of PLB-Thr-17. In contrast but not in those from patients with AF. In addition, KN-93 re- tothetrabeculaefrozenintheoperating theater, phosphory- duced the NE-evoked ICa,L responses in myocytes from patients lation in trabeculae after contracting for 1 h in an organ bath with SR but not in those from patients with AF, consistent with were reduced to only marginal signals at Ser-16 and Thr-17 a loss in AF of the NE-facilitated phosphorylation of the L-type + (SI Appendix,Fig.S8). Ca2 channel by CaMKII (Fig. 5 A and B). In contrast to a previous report (11), here the phosphorylation of PLB at Ser-16 and Thr-17 was consistently lower in trabeculae Reduced Agonist-Evoked Phosphorylation of PLB by CaMKII, but No frozen in the operating theater from patients with AF than in Change in RyR2 Phosphorylation Catalyzed by CaMKII and PKA in AF. those from patients with SR. The high basal phosphorylation NE, EPI, and FSK induced the phosphorylation of PLB at Thr- levels in trabeculae frozen in the operating theater likely reflect 17 in contracting trabeculae of both patients with SR and patients catecholamine surges during surgery. After contracting for 1 h in with AF, but to a lesser degree in the latter AF (Fig. 5C and SI the organ bath, the endogenous catecholamines dissipated from Appendix,Fig.S7). The effects of 5-HT were small in patients with the trabeculae, as revealed by the marginal phosphorylation SR, but even smaller in those with AF (Fig. 5C and SI Appendix, levels. The decreased phosphorylation of basal PLB at Ser-16 Fig. S7B). Increased phosphorylation of RyR2 channels at Ser- and Thr-17 in AF correspond well to the rightward shift in the 2808, catalyzed by PKA (11, 12), and Ser-2814, catalyzed by CaMKII, has been implicated in the arrhythmogenic release of concentration-effect curves for agonists and FSK in trabeculae of + patients with AF (Fig. 1 and SI Appendix, Table S2). The has- Ca2 through leaky RyR2 channels (13, 20); however, in contrast tening of trabecular relaxation caused by 10 μM of the agonists to previous reports, neither basal phosphorylation (SI Appendix, and FSK did not differ between patients with SR and those with Figs. S6 and S7) nor phosphorylation stimulated by agonists of AF (SI Appendix, Fig. S9). Accordingly, the agonists and FSK RyR2 at Ser-2808 (Fig. 4B and SI Appendix, Fig. S7A) and Ser- increased phosphorylation of PLB-Ser-16, Tn-I-Ser-23/24, and 2814 (Fig. 5C and SI Appendix, Fig. S7B) were different between Prot-C-Ser-282 similarly in trabeculae of patients with SR and patients with SR and those with AF. patients with AF (Fig. 4B and SI Appendix, Fig. S7). In both Discussion groups, stimulation of PLB-Ser-16 by 5-HT was less pronounced than stimulation by NE, EPI, or FSK. Our results are consistent Surprisingly, we hardly detected arrhythmic contractions in atrial with an early report indicating that ISO-evoked relaxation and trabeculae of the patients with AF. This finding is inconsistent with the idea of leaky sarcoplasmic reticulum in AF. The agonists AC stimulation do not differ between patients with SR and + must have produced an additional Ca2 load, as has been reported patients with AF (18). + under βAR stimulation (21). The increased Ca2 load facilitates KN-93 Reduces NE-Evoked Increases of ICa,L in Patients with SR, but the release process (22), consistent with the increased CaTs ob- Not in Patients with AF. We next compared the role of CaMKII on served with 5-HT, FSK, and ISO in cardiomyocytes from patients 2+ basal ICa,L and the activation of ICa,L by NE in myocytes from with SR. βAR stimulation increases the Ca sensitivity of the patients with SR and patients with AF. As reported previously RyR2 channels and reportedly increases the likelihood of RyR2 (13, 19), the CaMKII inhibitor KN-93, but not its inactive analog clusters opening and triggering release from neighboring clusters

11196 | www.pnas.org/cgi/doi/10.1073/pnas.1324132111 Christ et al. (22). CaTs are reportedly biphasic at room temperature (23) in preceding PKA-catalyzed phosphorylation of PLB at Ser-16, + the absence of agonists in human atrial myocytes. The early phase would increase the Ca2 load of the sarcoplasmic reticulum. 2+ 2+ 2+ is apparently due to Ca -induced Ca release by ICa,L from the SERCA’sCa pumping activity then may be further enhanced longitudinal junctional sarcoplasmic reticulum. The late phase by phosphorylation of PLB at Thr-17 (29), resulting in additional + + appears to occur through Ca2 release from corbular reticulum, Ca2 overload of the sarcoplasmic reticulum and release and + by Ca2 released from neighboring RyR2 channels, independent thus facilitating the appearance of SDCRs and arrhythmias ob- of ICa,L, as suggested by recent modeling (24). In myocytes from served with agonists and FSK. + our patients with SR, Ca2 transients were monophasic at 37 °C; Increased CaMKII-catalyzed phosphorylation of PLB and however, a late phase became conspicuous with 5-HT, FSK, and RyR2 in the absence of agonists have been reported in AF and ISO at 37 °C, suggesting that these agents cause propagation of implicated in the triggering and/or maintenance of this arrhyth- + Ca2 release from the periphery into the myocyte interior through mia in human atrium (11, 13, 20). The reports of CaMKII-cata- collection of RyR2 channels, resulting in increased contractility lyzed hyperphosphorylation have suggested the therapeutic use of and propensity to arrhythmias. CaMKII inhibitors in AF (13, 30); however, our evidence indi- Arrhythmogenic, delayed afterdepolarizations (DADs) have cates that CaMKII-dependent phosphorylation is either reduced + been reported with ISO via spontaneous diastolic Ca2 releases, or unchanged in human AF. Phosphorylation of PLB at Thr-17 with a subsequent increase in inward depolarizing current in- in trabeculae in the operating theater (SI Appendix, Fig. S8)and + + duced by the Na -Ca2 exchanger (25), which in turn could elicit in contracting atria by the agonists and FSK (Fig. 5) was reduced in DADs. 5-HT, FSK, and ISO caused SDCRs in myocytes of AF. These findings, together with the loss of the CaMKII-facil- patients with SR. In patients with AF, SDCRs were absent with itated component of both basal ICa,L and agonist-evoked ICa,L 5-HT and reduced with FSK, suggesting that the blunting of response, may contribute to the decrease in arrhythmias. Phos- arrhythmias elicited by 5-HT and FSK in AF are related, at least phorylation by CaMKII of RyR2 at Ser-2814 by the agonists and in part, to the reduction in SDCRs. FSK was unchanged in patients with AF compared with patients + CaMKII participates in the generation of catecholamine- with SR (Fig. 5), suggesting unchanged modulation of Ca2 re- evoked arrhythmias in animal models. Arrhythmogenic effects lease through this RyR2 phosphorylation site. of ISO were found to be antagonized by CaMKII inhibitors in The marked blunting or disappearance of atrial arrhythmias + rabbit ventricle (26). Increased Ca2 leaking and DADs have elicited by the agonists and FSK in AF also could be related in been reported in myocytes from mice overexpressing CaMKII part to the more negative resting membrane potential. This hy- and ISO elicited ventricular arrhythmias that were prevented by perpolarization is caused by a twofold increase in the inward KN-93 (27). Particularly relevant to AF, KN-93 prevented ISO- rectifying current (IK1) reported in myocytes of patients with AF evoked arrhythmic activity in rabbit pulmonary veins (28). compared with myocytes of patients with SR (31). Hyperpolar- We have demonstrated that the agonists and FSK caused PLB ization in AF would reduce the likelihood of depolarizing cur- phosphorylation at Thr-17 by CaMKII, conceivably activated by rents reaching the threshold for arrhythmogenic depolarizations. 2+ preceding increased cytoplasmic Ca levels owing to PKA- and The maximum response of ICa,L to the catecholamines, 5-HT, 2+ CaMKII-mediated increased ICa,L, as well as to Ca release and FSK was similarly reduced by one-third in the patients with from the sarcoplasmic reticulum (29). SERCA, activated by the AF. But although the inotropic response to the catecholamines

SRSR AF AF A 0 2 min PHARMACOLOGY

10 pA/pF

KN-92 KN-93 KN-92 KN-93 NE NE NE NE p < 0.001 B Fig. 5. Evidence for decreased function of CaMKII in AF. (A) ### 20 20 Time course of ICa,L amplitude in two myocytes from patients with SR (Left) and in two myocytes from patients with AF 15 15 (Right), representing the decreases in both basal ICa,L and NE- ### ### 10 10 ### evoked response by the active CaMKII inhibitor KN-93 (pA/pF) (20 μM) in patients with SR, but not in patients with AF. KN- Ca,L I * 5 5 92 (20 μM) was ineffective. The dotted line indicates a cur- 12/6 24/9 9/6 23/13 rent level of 0. (B) Data summary from several experiments. 0 0 ### B+KN-92 NEKN-93B+ NEB+KN-92 NEKN-93B+ NE P < 0.001, NE in the presence of KN-92 or KN-93, re- C spectively. *P < 0.05, KN-93 vs. the corresponding basal. (B) Thr-17 2814 Numbers of experiments are given in cells/patients. (C) Western blots of CaMKII-catalyzed phosphorylation by ago- CSQ nists and FSK of PLB-Thr-17 (Left) and RyR2-Ser-2814 (Right). SR (n=12) Same design as in Fig. 4B. NE, EPI, and FSK caused similar PLB-Thr-17 0.15 RyR2-Ser-2814 SR (n=12) 2 AF (n=11) AF (n=12) increases of phosphorylation of PLB-Thr-17, whereas the

) responses to 5-HT were smaller. The phosphorylation of PLB 5 1.5 # 0.1 10 at Thr-17 was reduced in patients with AF compared with # 1 # # patients with SR. **P < 0.01; ***P < 0.001, 5-HT vs. FSK. P <

a.u. ( 0.05 0.05, AF vs. SR; P = 0.06. AF vs. SR for 5-HT. The RyR2 phos- 0.5 ** phorylation at Ser-2814 by the agonists and FSK were not *** 0 significantly different in trabeculae from patients with SR EPINE 5-HT FSK EPINE 5-HT FSK EPINE 5-HT FSK EPI NE 5-HT FSK and trabeculae from patients with AF.

Christ et al. PNAS | July 29, 2014 | vol. 111 | no. 30 | 11197 and FSK was similar in patients with SR and those with AF, it , inconsistent with a contribution of endogenous was markedly reduced for 5-HT in patients with AF, suggesting catecholamines. Our work identifies mechanisms for the vanishing + uncoupling between the L-type Ca2 current and contractile of catecholamine-evoked AF in atrium that has undergone processes. The low 5-HT4R density compared with the density of remodeling. CaMKII appears to facilitate agonist-evoked arrhyth- β ∼ 1AR (32) and the 100-fold smaller sarcolemmal cAMP/PKA mias in atrial myocardium of patients with SR; however, once signals for 5-HT compared with NE and EPI (Fig. 4A) already chronic AF is established, agonists fail to elicit arrhythmias, prob- produce lower functional and biochemical signals in patients ably related to atrial remodeling, which includes decreases in SI Appendix with SR (Figs. 4 and 5 and , Fig. S9). In patients with CaMKII-mediated processes. Because of this pathophysiological AF, the difference between reduced 5-HT signals and cate- adaptation, it appears redundant to use CaMKII inhibitors in cholamine signals became more pronounced (Figs. 4 and 5), AF, as has been proposed by others. possibly related to the well-known increase in atrial myocyte size in AF (3, 4). This would increase the diffusion path and dissi- Materials and Methods pation of cAMP from its formation by the -bound Right atrial appendages were collected with informed consent from patients AC to its intracellular effectors. The especially small amounts of A undergoing cardiac surgery at the Department of Heart Surgery, Dresden cAMP produced by 5-HT (Fig. 4 ) would be expected to reach University of Technology. These studies were approved by the Medical Faculty these effectors at even lower concentrations in myocytes from Ethics Committee of Dresden University of Technology (document EK790799). patients with AF than in myocytes from patients with SR, thereby Patient characteristics are detailed in SI Appendix,TableS1. The methodology for reducing the inotropically relevant phosphorylations catalyzed experiments with atrial trabeculae, ICa,L, CaTs, and Western blots are described in by PKA. SI Appendix, Materials and Methods. Data comparisons were made with ANOVA Our evidence from human right atrium is a model for and nonpaired or paired t tests as appropriate, using GraphPad Prism and IBM arrhythmias caused by endogenous catecholamines and 5-HT. SPSS 19.0.1. Arrhythmias of this sort may trigger extrasystoles that could lead to paroxysmal AF or transient postoperative AF, the latter ACKNOWLEDGMENTS. We thank Dr. Thomas Eschenhagen for reading the of which is often prevented by treatment with βAR blockers (10). manuscript and Ms. Romy Kempe, Annett Opitz, Trautlinde Thurm, and Anne- β gret Häntzschel for providing valuable technical assistance. This work was sup- In contrast, AR blockers neither affect the incidence of chronic ported by the German Center of Cardiovascular Research. N.R. and U.R. AF (33) nor convert AF into SR (34) in patients with advanced received financial support from Fondation Leducq (Grant 07 CVD 03).

+ 1. Lloyd-Jones DM, et al. (2004) Lifetime risk for development of atrial fibrillation: The 20. Voigt N, et al. (2012) Enhanced sarcoplasmic reticulum Ca2 leak and increased + + Framingham Heart Study. Circulation 110(9):1042–1046. Na –Ca2 exchanger function underlie delayed afterdepolarizations in patients 2. Manning WJ, et al. (1994) Impaired left atrial mechanical function after cardioversion: with chronic atrial fibrillation. Circulation 125(17):2059–2070. + + Relation to the duration of atrial fibrillation. J Am Coll Cardiol 23(7):1535–1540. 21. Hussain M, Orchard CH (1997) Sarcoplasmic reticulum Ca2 content, L-type Ca2 cur- + 3. Schotten U, et al. (2001) Cellular mechanisms of depressed atrial contractility in pa- rent, and the Ca2 transient in rat myocytes during β-adrenergic stimulation. J Physiol tients with chronic atrial fibrillation. Circulation 103(5):691–698. 505(Pt 2):385–402. 4. Van Wagoner DR, et al. (1999) Atrial L-type Ca2+ currents and human atrial fibrilla- 22. Ogrodnik J, Niggli E (2010) Increased Ca2+ leak and spatiotemporal coherence of + tion. Circ Res 85(5):428–436. Ca2 release in cardiomyocytes during beta-adrenergic stimulation. J Physiol 5. Molenaar P, et al. (2007) (-)- elicits positive inotropic, lusitropic, and bio- 588(Pt 1):225–242.

chemical effects through β2-adrenoceptors in human atrial myocardium from non- 23. Hatem SN, et al. (1997) Different compartments of sarcoplasmic reticulum participate failing and failing hearts, consistent with Gs coupling but not with Gi coupling. in the excitation-contraction coupling process in human atrial myocytes. Circ Res Naunyn Schmiedebergs Arch Pharmacol 375(1):11–28. 80(3):345–353. 6. Kaumann AJ, Sanders L, Brown AM, Murray KJ, Brown MJ (1990) A 5-hydroxytryp- 24. Koivumäki JT, Korhonen T, Tavi P (2011) Impact of sarcoplasmic reticulum calcium re- tamine receptor in human atrium. Br J Pharmacol 100(4):879–885. lease on calcium dynamics and morphology in human atrial myocytes: A 7. Gergs U, et al. (2009) Phosphorylation of phospholamban and troponin I through computational study. PLOS Comput Biol 7(1):e1001067.

5-HT4 receptors in the isolated human atrium. Naunyn Schmiedebergs Arch Phar- 25. Wei SK, Hanlon SU, Haigney MC (2002) Beta-adrenergic stimulation of pig myocytes macol 379(4):349–359. with decreased cytosolic free magnesium prolongs the action potential and enhances

8. Kaumann AJ, Sanders L (1993) Both β1- and β2-adrenoceptors mediate catecholamine- triggered activity. J Cardiovasc Electrophysiol 13(6):587–592. evoked arrhythmias in isolated human right atrium. Naunyn Schmiedebergs Arch 26. Curran J, Hinton MJ, Ríos E, Bers DM, Shannon TR (2007) Beta-adrenergic enhance- Pharmacol 348(5):536–540. ment of sarcoplasmic reticulum calcium leak in cardiac myocytes is mediated by cal- 9. Kaumann AJ, Sanders L (1994) 5-Hydroxytryptamine causes rate-dependent ar- cium/calmodulin-dependent protein kinase. Circ Res 100(3):391–398. rhythmias through 5-HT4 receptors in human atrium: Facilitation by chronic β-adre- 27. Sag CM, et al. (2009) Calcium/calmodulin-dependent protein kinase II contributes to noceptor blockade. Naunyn Schmiedebergs Arch Pharmacol 349(4):331–337. cardiac arrhythmogenesis in heart failure. Circ Heart Fail 2(6):664–675. 10. Kalman JM, et al. (1995) Atrial fibrillation after coronary artery bypass grafting is 28. Lo L-W, et al. (2007) Calmodulin kinase II inhibition prevents arrhythmic activity in- associated with sympathetic activation. Ann Thorac Surg 60(6):1709–1715. duced by alpha and beta adrenergic agonists in rabbit pulmonary veins. Eur J Phar- + 11. El-Armouche A, et al. (2006) Molecular determinants of altered Ca2 handling in macol 571(2-3):197–208. human chronic atrial fibrillation. Circulation 114(7):670–680. 29. Grimm M, Brown JH (2010) Beta-adrenergic receptor signaling in the heart: Role of 12. Vest JA, et al. (2005) Defective cardiac regulation during atrial CamKII. J Mol Cell Cardiol 48(2):322–330. fibrillation. Circulation 111(16):2025–2032. 30. Rokita AG, Anderson ME (2012) New therapeutic targets in cardiology: Ar- + + 13. Neef S, et al. (2010) CaMKII-dependent diastolic SR Ca2 leak and elevated diastolic rhythmias and Ca2 /calmodulin-dependent kinase II (CaMKII). Circulation 126(17): Ca2+ levels in right atrial myocardium of patients with atrial fibrillation. Circ Res 2125–2139. 106(6):1134–1144. 31. Dobrev D, et al. (2001) Molecular basis of down-regulation of G-protein–coupled + 14. Seamon KB, Padgett W, Daly JW (1981) Forskolin: Unique diterpene activator of ad- inward rectifying K current (IK,Ach) in chronic human atrial fibrillation: Decrease in

enylate cyclase in membranes and in intact cells. Proc Natl Acad Sci USA 78(6):3363–3367. GIRK4 mRNA correlates with reduced IK,ACh and muscarinic receptor-mediated 15. Pau D, Workman AJ, Kane KA, Rankin AC (2007) Electrophysiological and arrhythmogenic shortening of action potentials. Circulation 104:2551–2557.

effects of 5-hydroxytryptamine on human atrial cells are reduced in atrial fibrillation. JMol 32. Kaumann AJ, Lynham JA, Brown AM (1996) Comparison of the densities of 5-HT4

Cell Cardiol 42(1):54–62. receptors, β1- and β2-adrenoceptors in human atrium: Functional implications. Nau- 16. Bers DM (2002) Cardiac excitation-contraction coupling. Nature 415(6868):198–205. nyn Schmiedebergs Arch Pharmacol 353(5):592–595. 17. Van Haastert PJ, et al. (1984) Competitive cAMP antagonists for cAMP-receptor 33. Swedberg K, et al. (2005) Prognostic relevance of atrial fibrillation in patients with proteins. J Biol Chem 259(16):10020–10024. chronic heart failure on long-term treatment with beta-blockers: Results from 18. Schotten U, et al. (2002) Atrial fibrillation-induced atrial contractile dysfunction: A COMET. Eur Heart J 26(13):1303–1308. tachycardiomyopathy of a different sort. Cardiovasc Res 53(1):192–201. 34. Kao DP, et al. (2013) Effect of bucindolol on heart failure outcomes and heart rate 19. Christ T, et al. (2004) L-type Ca2+ current down-regulation in chronic human atrial fibrillation response in patients with reduced ejection fraction heart failure and atrial fibrillation. is associated with increased activity of protein phosphatases. Circulation 110(17):2651–2657. Eur J Heart Fail 15(3):324–333.

11198 | www.pnas.org/cgi/doi/10.1073/pnas.1324132111 Christ et al.